AVOIDING CAVERNS IN THE ARBUCKLE MOUNTAINS USING ELECTRICAL IMAGING METHODS

Peter J. Hutchinson THG Geophysics, Ltd., 4280 Old Wm. Penn Hwy., Murrysville, Pennsylvania, 15668,USA, [email protected]

Abstract farms. As of 2016, has 3,134 megawatts The Arbuckle Mountains in Oklahoma are a unique of installed wind capacity (AWEA, 2017). Monies province with nearly vertical dipping beds of the Upper (2017) reported that as of 2017, Oklahoma had 6,600 Cambrian to Lower Ordovician Arbuckle Group. The megawatts of wind power that supply one quarter of Arbuckle Group consists of intercalated sequences of Oklahoma’s generated electricity. thick carbonates and thinner shales and . Syn- mountain building steeply tilted these The permitting process for a wind farm provides beds. Subsequently, caves and voids developed within the limited tower relocation potential without an carbonate beds, presumably by hypogene speleogenesis. extensive repermitting process; however, most permits provide for some limited relocation of Numerous dissolution cavities and several major cave a tower within an approximately 30 m radius of systems have been mapped within the Arbuckle Mountain the tower without going through the repermitting Wind Farm located 19 km north of Ardmore, Oklahoma. process. Wind turbine towers consist, from the Electrical resistivity imaging was determined to be the most ground up, of the foundation, stalk, nacelle, and rotor effective method of mapping subsurface voids and caverns blades. The foundation, in general, is a 15 m wide by due to the strong electrical contrast between carbonate units 3 m deep steel-reinforced concrete subsurface slab. and void-filling sediments and groundwater. Unstable foundations can cause the towers to topple so the location of a stable foundation is critical to the Electrical resistivity imaging during early stages of the installation of a tower. wind farm development identified numerous caverns and voids beneath proposed turbine sites; consequently, The wind turbine towers are erected by crawler cranes several wind tower locations were moved. This report that are slow (1.4 km/hr), 187 m tall, and very heavy addresses the East Access Road to six proposed turbine with a ballast weight of approximately 600 t (Liebherr, wind tower sites where three towers and access road 2017). Crawler cranes are erected at a staging area near the towers were relocated south of their original and crawl to the individual towers for construction proposed locations. Further, the Main Access Road of the wind turbine towers. Further, crawler cranes to these six tower sites had to be rerouted due to the due to their height can only tip up to 5° from vertical presence of two previously unidentified sinkholes and a before catastrophic collapse. Consequently, the road major cavern system, the Wild Woman Cave Complex. upon which the crane travels must be stable.

Electrical imaging method identified subsurface Areas with poor agricultural potential create ideal anomalies along the proposed access roads and at four opportunities for wind power development, since the of the six towers. The towers were moved away from the land is often remote and underutilized. One of these subsurface anomalies and the access roads were relocated areas is located in the southern portion of the Arbuckle to positions where subsurface anomalies would not pose Mountains, eponymously named after General a hazard to the heavy crawler cranes, used to erect the Matthew Arbuckle (1778–1851). The 6,000-hectare towers. The 50-tower Anadarko Mountain Wind Farm Arbuckle Mountain Wind Farm consists of 50 was successfully completed in 2016. wind turbine generators that can produce 100 MW of electricity. Construction of the wind farm was Introduction challenging due to the presence of topography Oklahoma ranks 8th for wind energy potential and within the steeply dipping beds of and maintains over 2,000 wind turbines in over 27 wind dolostone sediments (Figure 1).

15TH SINKHOLE CONFERENCE NCKRI SYMPOSIUM 7 347 Caverns and passages at the site have been mapped by The predominant rocks of interest at the site are the karst- various grotto groups for decades and provided a useful forming and dolostones. These carbonate characterization of the subsurface anomalies in the study rocks are estimated to be many thousands of feet thick area. For example, Harrel (1959) published a detailed (Ham, 1973). Surficial karst features observed in the cave map of Wild Woman Cave (Figure 2). field consisted of solution weathered vugs, bedding plane fractures, and voids. Geology Arbuckle Mountain Wind Farm is situated in the western Three geological groups lie within the footprint of the portion of the Arbuckle Mountains of Arbuckle Wind Project. They are listed in order of (Figure 1). This portion of the Arbuckle Mountains is decreasing age: middle Cambrian Colbert ; noted for having karst topography. The rock units consist upper Cambrian/lower Ordovician Arbuckle Group; and predominantly of limestone, dolomite and rhyolite, with middle Ordovician Simpson Group. minor amounts of shale and . The carbonate units are characterized by having sinkholes, dolines, and caverns. The Colbert Rhyolite, located in the northern portion of wind farm, is a “…pink porphyritic rhyolite, locally The rocks in this portion of the Arbuckle Mountains flow-banded…” (Ham, 1951). are part of the Arbuckle and are contiguous with a south-dipping thrust fault, the Washita Valley The Arbuckle Group contains 8 Formations, the Fort Fault Zone (Johnson, 1990). The sediments at the Sill Limestone, Royer Dolomite, Signal Mountain wind farm site strike N45°W and dip up to 45°SW. Formation, Butterly Dolomite, McKenzie Hill Formation, Deformation or mountain building occurred during Cook Creek Formation, Kindblade Formation, and the the Middle Pennsylvanian time and ended during the West Spring Creek Formation. These formations are Late Pennsylvanian as flat-lying upper Pennsylvanian composed of limestone and dolomite with a maximum and rocks overly the steeply inclined lower thickness of 2,000 m in the Arbuckle Mountain region Pennsylvanian rocks in other areas nearby. (Ham, 1973). Thin sandstone beds are present within the

Figure 1. Geologic map of the western portion of the Arbuckle Mountains showing the locations of proposed wind turbine towers (green dots) to the Arbuckle Mountain Wind Farm (Johnson, 1990).

348 NCKRI SYMPOSIUM 7 15TH SINKHOLE CONFERENCE Figure 2. Subsurface plan map of Wild Woman Cave, Arbuckle Mountains (Harrel, 1959).

West Spring and Cool Creek Formations and cherty and Where, resistance, R (Ohms), is equal to the ratio sandy limestone is found throughout the Cool Creek and of potential, V (volts), to current flow, I (amperes). McKenzie Hill Formations. Resistivity, then, is a measure of the resistance (ρ) along a linear distance (L) of a material with a known cross- The Simpson Group contains 5 Formations, the Joins, Oil sectional area (A). Creek, McLish, Tulip Creek, and Bromide Formations. R = ρL / A;where V A These units have a maximum thickness of 700 m in the where ρ = Arbuckle Mountain region and are mainly comprised I L of limestone and dolomites with basal sandstones and Consequently, resistivity is measured in Ohm-meters minor amounts of shale (Ham, 1973). and the data can be presented as a geo-electrical profile of modeled apparent resistivity versus depth. During the formation of dolomite the net rock volume of limestone decreases leaving voids and vugs that can Electrical currents propagate as a function of three induce further dissolution. Klimchouk (2007; 2014) material properties: (1) ohmic conductivity, (2) documents hypogene speleogenesis, or vertically electrolytic conductivity, and (3) dielectric conductivity. upward migration of groundwater during the course of Ohmic conductivity is a property exhibited by metals depositional history, as a mechanism for the creation (Kaufman, 1992). Electrolytic conductivity is a function of karst topography. Eschberger (2012) and Eschberger of the concentration of total dissolved solids and salts et al. (2014) noted hydrothermal intrusion during the in the groundwater that exists in the pore spaces of a Cambrian. Remanent magnetization of ferro-magnesium material (Reynolds, 1997). Dielectric conductivity is a minerals during Pennsylvanian–Permian deformation function of the permittivity of the matrix of the material has been attributed to the percolation of orogenic fluids (von Hippel, 1954). Therefore, the matrix of most soil through the carbonate units (Nick and Elmore, 1990). and bedrock is highly resistive. Of these three properties, Sykes (1997) documented active basinal fluids during electrolytic conductivity is the dominant material the Arbuckle and Laramide . Vertically upward characteristic that influences the apparent resistivity flushing has continued to the present (for example, see values collected by this method (Milsom, 1989). Puckett, 2009; and Christensen et al., 2011). Blackwood (2017) and Blackwood et al. (2015) argued successfully In general, resistivity values decrease in water-bearing that hypogenic flow created or initiated the dissolution rocks and soil with increasing: fractional volume of the processes, or speleogenesis, within the Arbuckle rock occupied by groundwater; total dissolved solid and Mountains. chloride content of the groundwater; permeability of the pore spaces; and, temperature. Theory and Methods Electrical resistance is based upon Ohm’s Law: Materials with minimal primary pore space (i.e., R=V/I limestone, ) or lack groundwater in the pore

15TH SINKHOLE CONFERENCE NCKRI SYMPOSIUM 7 349 spaces will exhibit high resistivity values (Mooney, of each block are then calculated to produce an apparent 1984). Highly porous, moist or saturated soil, such as fat resistivity pseudosection. The pseudosection is compared clays, will exhibit very low resistivity values. to the actual measurements for consistency. A measure of the difference is given by the root-mean-squared (rms) In homogeneous ground, the apparent resistivity is the true error. ground resistivity; however, in heterogeneous ground, the apparent resistivity represents a weighted average of all Discussion formations through which the current passes (Mooney, Since air (i.e., voids) and carbonate rock have very 1984). Many electrode placements (arrays) have been high apparent resistivity values it is difficult to proposed (for examples, see Reynolds, 1997), including distinguish between the two. Fortunately, most of the the most commonly deployed Wenner, Schlumberger, and voids at the site, or at least those we could identify, dipole-dipole array. Two survey arrays, dipole-dipole and contained clay and/or water; both of which have very Schlumberger, were collected for each tomograph. low apparent resistivity values. For this site then, low apparent resistivity values and inversions (i.e., resistive The Schlumberger array has proven to be an effective material overlying conductive material) are considered configuration for imaging the subsurface: to be indicators of weak rock conditions at depth with the possibility of the presence of vuggy subsurface Ri = π a2/b[1 – b2/4 a2] R; a = 5b conditions or voids.

Where, resistivity (Ri) is related the separation distance Initially, paired electrical imaging profiles were collected between the current source and current sink (b), and the at all 50 tower sites, which included (at the client’s pole spacing, (a). request) the three sites located on the Colbert Rhyolite (Figure 1). The possibility of a void in the Colbert, The dipole-dipole array collects a denser, deeper portion however, is remote and no subsurface anomalies were of the subsurface, and combined with the Schlumberger detected. The survey consisted of a pair of 69 m long array, provides data for the construction of a detailed lines set at right angles to each other. Each profile was geo-electric profile. Dipole-dipole configuration is: centered at the staked locations of the proposed tower and imaged the subsurface geo-electric stratigraphy to a R = π a n(n + 1)(n + 2) depth of approximately 18 m below grade.

A forward modeling subroutine was used to calculate Eighteen sites were determined to have potential the apparent resistivity values for dipole-dipole and subsurface anomalies (Table 1). All were relocated away Schlumberger data sets, after which the data sets were from the subsurface anomalies and the new locations combined using the EarthImager program (AGI, 2002). drilled to insure stable subsurface conditions. However, This program is based on the smoothness-constrained during this work, it was noted that the Main Access least-squares method (deGroot-Hedlin and Constable, Road to Towers 38 through 43 crossed Wild Woman 1990; Loke and Barker, 1996). The smoothness- Cave Complex (Figure 3). This report documents the constrained least-squares method is based upon the additional work that was performed to relocate the Main following equation: Access, East Access, and West Access Roads (Figures 4, 5, 6, and 7). JT g = (JTJ+ μF)d During the course of imaging the foundations at towers Where, F is a function of the horizontal and vertical 38 through 43, several subsurface anomalies were noted flatness filter (J) is the matrix of partial derivatives, μ is (Table 1). Further, this area was considered to be a the damping factor, d is the model perturbation vector potential hazard for the crane. Profile 1 and 1A through and g is the discrepancy vector. 1F; Profiles 2A and 2B; and Profiles 3A and 3B were collected to assess the potential hazard to the heavy The EarthImager program divides the subsurface 2-D crane along the East and West Access Roads (Figures 4, space into a number of rectangular blocks. Resistivities 5, and 6, respectively).

350 NCKRI SYMPOSIUM 7 15TH SINKHOLE CONFERENCE Turbine Formation From the previous study, subsurface anomalies were 16 noted at Towers 41, 42, and 43 (Table 1). Profile 1 and 17 cross lines 1A through 1F show the presence of numerous 22 subsurface anomalies along the proposed crane path 26 (Figure 4). Due to the potential risk to the the towers 27 West Spring and Kindblade and crane, Towers 41, 42, and 43 and the portion of the 28 East Access Road between the towers were relocated 29 approximately 20 m to the south (Figure 8). 30 31 The southern portion of the East Access Road was inferred 33 to cross Persimmon Cave. Two profiles; Profile 2A and 2B, 40 were collected in the area to determine if any subsurface 41 anomalies existed (Figure 5). Persimmon Cave is inferred to 42 Cool Creek and McKenzie Hill exist at depth at position 460 ft (140 m) on Profile 2A. There 43 is approximately 8 m of interpreted hard rock overlying the 50 inferred cave (Figure 5). Due to the interpreted size and 47 depth of the cave, the proposed location of the East Access 48 Butterly, Signal Mountain, Royer & Fort Sill Road that crossed over the inferred location of Persimmon 49 Cave was not interpreted to be a hazard to the crane and the road was not relocated (Figure 8). Table 1. Table of towers with EI identified anomalies.

Figure 3. Topographic map (contour interval 2 feet; 0.6 m) showing the originally proposed roads and the electrical imaging profile locations. The location of the Wild Woman Complex is from Curtis (1959) and the sinkhole locations are from Blackwood (2014). Location of Persimmon Cave is approximated and is not based upon real information. Map scale is set to Oklahoma State plane coordinate system, NAD 1983, in feet.

15TH SINKHOLE CONFERENCE NCKRI SYMPOSIUM 7 351 Figure 4. Profile Line 1 and cross Profile Lines 1A through 1F. Color scale shown is used on all figures but only reproduced here. Profiles have no vertical exaggeration. Subsurface anomalies are noted with dashed black ovals.

Figure 5. Profile Line 2A and 2B. Color scale from Figure 4. Profiles have no vertical exaggeration. Subsurface anomalies are noted with dashed circular black lines.

352 NCKRI SYMPOSIUM 7 15TH SINKHOLE CONFERENCE Figure 6. A portion of Profile Line 3 and all of Profile Line 3B. Color scale from Figure 4. Profiles have no vertical exaggeration. Subsurface anomalies are noted with dashed circular black lines.

Figure 7. Portion of Profile Line 4A, 4B, and 4C. Color scale from Figure 4. Profiles have no vertical exaggeration. Subsurface anomalies are noted with dashed circular black lines.

15TH SINKHOLE CONFERENCE NCKRI SYMPOSIUM 7 353 Further south along the East Access Road to Towers 41 grade (Figure 6). Due to its interpreted size and depth, through 43, two subsurface anomalies were determined this anomaly was not considered to be a hazard to the to be enough of a hazard to warrant relocation of the crane. It should be noted that Tower 40 was moved 10 m crane path (Figures 5 and 8). The crane path was moved to the west to avoid a subsurface anomaly. 6 to 10 m southeast of the proposed path to avoid any subsurface issues (Figure 8). The Main Access Road to the six towers crossed portions of Wild Woman Cave Complex, that necessitated the The West Access Road to Towers 38 through 40 showed relocation of the road (Figure 3). At the southern end of only one subsurface anomaly at a depth of 10 m below the road, Sinkhole #1 was discovered along the proposed

Figure 8. Aerial map of Anadarko Mountain Wind Farm showing tower locations (black dots) and as-built access roads (Google Earth, 2017).

354 NCKRI SYMPOSIUM 7 15TH SINKHOLE CONFERENCE road. Profile 4A at position 115 ft (35 m) shows an Degroot‐Hedlin C, Constable S. 1990. Occam’s anomaly inferred to be this sinkhole (Figure 7). Profile inversion to generate smooth, two‐dimensional 4A may have imaged Wild Woman Cave (position models from magnetotelluric data. Geophysics 55 2,050 ft; 630 m) at a depth of approximately 14 m below (12): 1613–1624. grade. Further, Profile 4B and 4C imaged Sinkhole #2 Elmore RD. 2001. A review of paleomagnetic data on at a depth of approximately 8 m below grade. A small the timing and origin of multiple fluid-flow events in the Arbuckle Mountains, Southern Oklahoma. anomaly interpreted to be a cavern is located on Profile Petroleum Geoscience 7 (3): 223–229. 4B at position 1,770 ft (540 m). Eschberger AM, Hanson RE, Puckett RE. 2014. Carlton Rhyolite Group and Diabase Intrusions in the Wild Woman Cave was not considered to be a hazard East Timbered Hills, Arbuckle Mountains. In: to the crane as the cave was interpreted to be located at Suneson NH editor. Oklahoma Geological Survey. a depth too deep to impact the crane. The Main Access Guidebook 38: 143–186. Road, now termed the New Access Road, was rerouted Eschberger AM. 2012. Volcanological and geochemical around Sinkhole #2. Further, the interpreted anomaly at studies of Cambrian -related igneous rocks the end of Profile 4B was interpreted to pose a hazard in the western Arbuckle Mountains, southern to the crane and the beginning of the East Access Road Oklahoma [Master’s Thesis]. Texas Christian was moved north approximately 10 m to avoid the void. University, Fort Worth, Texas. Google Earth. 2017. Google Earth aerial photographic program. San Francisco (CA). Google. The wind farm was installed successfully without any Ham WE. 1951. Geology and petrology of the Arbuckle delays or problems. Consequently, electrical imaging limestone in the southern Arbuckle Mountains, methods are an effective method of identifying Oklahoma [PhD Dissertation]. Yale Univ. subsurface anomalies that pose a risk to engineered Ham WE. 1973. Regional geology of the Arbuckle facilities. Mountains, Oklahoma. Oklahoma Geological Survey Special Publication 1. References Johnson KS. 1990. Geologic Map and Sections of the AGI. 2002. EarthImager Program. American Arbuckle Mountains, Oklahoma; Circular 91, Geosciences Inc., Austin Texas. Plate 1 of 2, Oklahoma Geologic Survey; revised AWEA. 2017. American Wind Energy Association: from Ham WE, Mckinley ME, et al. 1954. Washington, DC. Kaufman AA. 1992. Geophysical Field Theory and Blackwood KW. 2014. Wild Woman Cave: Preliminary Method: Gravitational, Electric, and Magnetic Spatial Analysis. Unpublished document. Fields. New York (NY): Academic Press. Blackwood KW, Halihan T, Beard K. 2015. Kharaka Y, Law L, Carothers W, Goerlitz D. 1986. Role Development and distribution of hypogene caves of organic species dissolved in formation water and paleokarst features in the Arbuckle Mountains from sedimentary basins in mineral diagenesis. of South Central Oklahoma, USA. Abs. AAPG In: Gautier DL, editor. Roles of organic matter Annual Convention, Denver, Colorado. in sediment diagenesis. Society of Economic Blackwood KW. 2017. Hypogenic Caves and Paleontologists and Mineralogists, Special Paleokarst of the Arbuckle Mountains, Oklahoma. Publication 38: 111–122. In: Klimchouk AB, Palmer A, De Waele J, Auler Klimchouk AB. 2007. Hypogene speleogenesis: A, Audra P, editors. Hypogene Karst Regions and hydrogeological and morphogenetic perspective. Caves of the World. Cave and Karst Systems of National Cave and Karst Research Institute, the World. Springer, Cham. p. 653–661. Carlsbad, New Mexico. p. 5–97. Christensen S, Osborn NI, Neel CR, Faith JR, Blome Klimchouk AB. 2014. The methodological strength of CD, Puckett J, Pantea MP. 2011. Hydrogeology the hydrogeological approach to distinguishing and simulation of groundwater flow in the hypogene speleogenesis. Hypogene Cave Arbuckle–Simpson aquifer, south-central Morphologies Special Publication. National Cave Oklahoma. U.S. Geological Survey Scientific and Karst Research Institute, Carlsbad, New Mexico. Investigations Report 2011–5029. p. 18–98. Liebherr. 2017. Crawler Crane Specifications. Liebherr- Harrel R. 1959. A preliminary report on the invertebrate International AG self-published brochure. animals of Wild Woman Cave. Proceedings of the Loke MH, Barker RD. 1996. Rapid least-squares inversion of Oklahoma Academy of Sciences, 40: 20–31. apparent resistivity pseudosections by a quasi-Newton

15TH SINKHOLE CONFERENCE NCKRI SYMPOSIUM 7 355 method. Geophysical Prospecting 44 (1): 131–152. Milsom J. 1989. Field Geophysics. New York (NY). Halstead Press. Monies P. 2017. Oklahoma governor signs bills to end wind incentive, airfield loophole. 18 April 2017. Oklahoma City (OK). The Oklahoman. Mooney HM. 1984. Handbook of Engineering Geophysics. Minneapolis (MN). Bison Instruments, Inc. Nick KE, Elmore RD. 1990. Paleomagnetism of the Cambrian Royer Dolomite and Pennsylvanian Collings Ranch Conglomerate, southern Oklahoma: an early Paleozoic magnetization and non-pervasive remagnetization by weathering. Geological Society of America Bulletin, V102. p. 1517–1525. Puckette JT, Halihan T, Faith J. 2009. Characterization of the Arbuckle-Simpson Aquifer-Final Report for the Arbuckle-Simpson hydrology study. Stillwater (OK). Oklahoma State University, School of Geology, for the Oklahoma Water Resources Board. Reynolds J. 1997. An introduction to applied and environmental geophysics. New York (NY). John Wiley & Sons. Sykes M. 1997. Paleokarst characteristics of the surface and subsurface in the Viola Limestone (Ordovician), Arbuckle Mountains. Shale Shaker. Oklahoma City Geological Society. Oklahoma (OK). 47: 107–121. Von Hippel AR. 1954. Theory. Von Hippel AR. Editor. Dielectric Materials and Applications. Cambridge (MA). The MIT Press.

356 NCKRI SYMPOSIUM 7 15TH SINKHOLE CONFERENCE